When a few computers are the only casualties of an office fire, we’re typically satisfied that all the employees escaped unharmed. In a large data center space, though, where people are scarce and the equipment is worth millions, the loss of even a few computers can be crippling.
And while just about any fire suppression system will likely safeguard the property it is assigned to protect, there’s a small array of systems applicable to today’s mission critical environment. The essential questions to ask when determining the right one for your facility are: How much “insurance” is required for the space? What is the tolerance for business interruption?
There are four different fire suppression systems that are typically used to protect a data center. From the water-based pre-action and water mist systems to clean agent and in-rack suppression, these fire suppression systems build on each other, providing sequential levels of protection, with the most commonly specified system providing a first tier level of defense to the most costly and complicated system delivering the highest level of coverage.
LEVEL I: THE PRE-ACTION SYSTEM
The most commonly used, water-based suppression method — the pre-action system — is the most inexpensive and uncomplicated of the four data center fire suppression systems. It’s relatively easy to install and won’t be foreign to contractors or maintenance personnel charged with installing and maintaining it. Providing a basic level of insurance aimed at protecting the data center’s equipment, the pre-action system is an integrated fire suppression and fire alarm system housed within the pre-action fire alarm panel.
Initially, before the water in the system is first activated, the sprinkler system pipes are dry, providing a small window of insurance. Water is electrically or pneumatically restrained by a solenoid valve until the system’s detection sensors recognize either heat or smoke in the space, at which time the water will be released into the dry pipes. Simply put, water will only be released from the pipes via the sprinklers if one or more of the sprinklers are activated by heat. Once activated, the water will dowse the flames, typically destroying everything in the area of activation, from the fire to the equipment.
Pros: Widely used for its relatively simple installation, ease of maintenance, and cost-effectiveness, the pre-action system is dry, providing an initial level of equipment protection. When designed and maintained properly, the pre-action system is accident proof, only activating when it detects real fire.
Cons: Use of the pre-action system requires a high tolerance for business interruption, which most data centers don’t have. If the sprinklers would fully discharge, all the equipment on the data center floor would be destroyed.
Case in point: A 100,000-sq-ft Midwest-based data center supporting the operations of a large bank opted to employ a pre-action system for their fire protection. Although the data center was built to support 24/7 operations, the bank already had an identical, redundant data center in another state, and therefore had more of a tolerance for business interruption. The bank was able to save on capital and overall maintenance costs by specifying a pre-action system.
LEVEL II: THE CLEAN AGENT SYSTEM
Similar to the pre-action model, the clean agent fire suppression system also discharges from orifices scattered across the data center’s ceiling upon detection of smoke or heat. The clean agent system is unique in that it uses a clean agent — fire suppressant chemical or inert gas — to put a fire out. Advantageous for data centers that have little or no tolerance for business interruption, as many of the clean agents available today won’t ruin the equipment, clean agent systems are relatively more complicated to design, requiring volumetric calculations to ensure the right clean agent saturation percentage to extinguish a fire.
Pros: With a clean agent system, the chemical response can do little to no damage to the actual equipment in the space, while still thoroughly extinguishing the fire (assuming the system was specified and installed correctly). There are a number of different clean agent types including inert gasses and other agents, like halon, that can be employed in a clean agent system. Halon, while it is no longer produced due to the Clean Air Act Amendment of the 1990s, can be reused and recycled where the most effective clean agent is desired.
Cons: Clean agent systems are relatively more complicated to design and install and therefore tend to carry a higher price tag than their pre-action predecessor. Because the success of the system depends heavily upon having the right chemical saturation percentage, make sure to retain a knowledgeable fire protection engineer and fire protection contractor who has experience with these systems. If not, the system’s ability to extinguish the fire can be compromised. Saturation percentages are determined by the system’s manufacturer and will be different for each chemical agent. Additionally, the clean agent system needs to be constantly maintained to ensure an optimal level of chemical within the system so that it’s ready for discharge. Often authorities having jurisdiction (AHJs) and local building codes will require an NFPA 13-compliant system in addition to a clean agent system, which can be a pre-action system or another water-based system at the owner’s discretion. Where this is the case, it is important to regularly service both systems as well as the interface between the two to make sure all elements are in good working order.
Case in point: The main switchboard station of one telecom company carries wire connections from the East Coast to the West Coast. In high-priority rooms where the company needed to maintain a higher level of protection, a clean agent system was specified. The local AHJ required full building fire protection as well as a redundant water-based system in the areas that had a clean agent system only. According to the AHJ, this rule exists for multiple reasons: 1) The AHJ is not responsible for, nor can they require, appropriate maintenance of the clean agent system; 2) If the fire event occurs for an extended period of time, there is a concern that the clean agent system will run out of fire suppressant chemicals; and 3) If the fire is so big it overpowers the clean agent system, the AHJ wants the added security of a backup, water-based system.
LEVEL III: THE WATER MIST SYSTEM
In automatic sprinkler systems (of which the pre-action system is one), a minimum, code-required amount of water flows from any one individual sprinkler head when in discharge mode. However, as discussed, such a discharge will ruin the equipment in the process of extinguishing the fire. The idea behind the water mist system is to atomize the water instead, turning it into a mist — similar to that of a humidifier. The discharged water mist completely surrounds the fire, replacing oxygen via evaporation, while the sum total quantity of water is significantly less than the pre-action system — totaling only a small fraction of the water used by a pre-action sprinkler system for the same application area. The hybrid water mist system that recently became commercially available utilizes both inert gas and water mist to extinguish fires. Given its recent inception, there are no NFPA standards addressing this type of system. The only design guideline that currently exists is a factory mutual approval standard (FM5580).
Pros: The water mist system minimizes equipment damage compared to the pre-action system, and because the agent used to extinguish the fire is water, there’s no significant supply cost or maintenance requirements present as with the clean agent system.
Cons: In order for the water mist system to work successfully, a tremendous amount of pressure must be applied. The equipment used to pressurize the water can be very expensive, therefore, water mist systems are typically installed only in individual, critical rooms, as specifying a water mist system for an entire facility or data center floor would be too costly.
Case in point: A downtown high-rise installed a water mist system exclusively in an individual critical data center space. Because it was the first one installed in the city, the project’s MEP engineers had to meet with the AHJ to prove it could provide the required protection. Introduced in the last 40 years, the water mist system hasn’t been employed much to date. More often than not, when specifying a water mist system, special approval from the local AHJ may be required.
LEVEL IV: IN-RACK PROTECTION
In-rack protection is exactly what it sounds like — an immediate, localized response inside the actual server rack to extinguish a fire. Here’s how it works: smoke detection housed within the server rack notifies the clean agent chemical, also housed within the server, to extinguish the fire. The dry, in-rack protection system minimizes equipment loss and business interruption. It offers the highest form of reliability out there today and is by far the most complex and costly of the four fire suppression systems. In-rack protection systems are growing in popularity and will be specified more often in the coming years.
Pros: The most significant advantage to specifying in-rack protection is the ability to contain the fire and its damage to only those racks that are affected. Even if a fire event occurs, without water, the in-rack suppression won’t ruin the equipment it is housed in, and will seek to eliminate the fire immediately since it is so localized.
Cons: Because the chemical agent is housed in the server racks, maintaining the appropriate level of chemical in each rack will require constant upkeep. Additionally, the local AHJ will likely require an automatic wet sprinkler or pre-action system to be installed at the ceiling to protect the rest of the data center floor. The additional system in a shared room with the in-rack system will require additional coordination by the specifier and installer. Furthermore, in order to maintain the effectiveness of the in-rack protection, cabinets must be fully enclosed to properly suppress the fire.
Case in point: A financial firm in a large city opted to implement in-rack fire suppression for all of their server racks housed in the IDF floor of their office building. Because of the damage containment benefits and decreased risk of business interruption, the firm opted to implement this system type for all of the server racks on the floor — approximately 50,000 sq ft of space. However, given the fact that the building was required to be considered fully sprinklered by the local building code, the AHJ required a pre-action system for the space as well.
As the data center owner or facility operator, your main goal is to safeguard the equipment inside your facility. Consider how critical that goal is to your organization, together with your on-site maintenance and operational capabilities, to determine how much money you’re willing to invest in fire suppression. Regardless of which level — or combination of — fire protection you choose, minimizing reaction time will be paramount to ensuring you are positioned to maintain your equipment long term.
Choose the detection system that best fits the occupancy.
Consider your tolerance to business interruption when choosing the detection system.
The more expensive the system, the more protection it will offer.
Thoroughly interface the detection system with the suppression system.